Personalizing Product Recommendations While Protecting Privacy

• Personalization offers significant benefits for brands, but consumer privacy concerns and evolving data protection regulations demand that algorithms are built on trust, transparency, and individual autonomy.
• Product recommendations that empower autonomy and align with user identities are less likely to feel coercive and trigger a response of psychological reactance.
• Transparency cues and privacy nudges offer consumers insight into the goals, purposes, and processes of personalization algorithms so they seem helpful rather than suspicious.
• Framing data collection as a mutually beneficial social exchange can ensure that the benefits of personalization outweigh the perceived privacy risks.

Personalization technology has significant potential to drive progress and promote positive behavior change. Among many other important applications, it can improve patient care, support self-led nutrition management, and transform classroom learning. But when it comes to product recommendations, personalization often feels more manipulative than meaningful. As personalized advertising becomes increasingly pervasive, a growing number of consumers are rejecting these marketing strategies, whether this means opting out of data sharing or installing ad-blocking software on their devices.^5,6 

Much of the time, these reactions are rooted in privacy concerns. Personalization—when done right—has the potential to increase customer loyalty, but it can have the opposite effect when consumers are concerned about privacy.⁷ Personalized product recommendations draw attention to the fact that user data is being collected and used, often beyond the user’s control. This intrusion erodes customer trust, and when that trust is broken, personalization strategies are more likely to drive customers away than help them make smart purchase decisions. 

The rise of data privacy laws like the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA) further complicates this issue by imposing restrictions on how data can be collected and used for recommendations. While crucial for ensuring the ethical use of data, half of businesses say recent legal changes have made personalization more difficult.4 What’s more, 88% of advertisers anticipate major changes to privacy laws that will impact personalized targeting.⁸ 

As companies grapple with changing consumer and regulatory demands, the question is no longer whether we can personalize product recommendations while complying with data regulations, but how we can do it in a way that delivers real value, earns trust, and respects user autonomy.

Even when personalization is technically helpful or relevant, it can trigger emotional resistance in users when it feels overly coercive or manipulative.⁶ We see this with personalization systems that make too many assumptions about user goals, for example, assuming that someone searching for hiking backpacks is gearing up for a major thru-hike and needs specialized gear recommendations. This can make users feel that the system is projecting an identity onto them instead of aligning with their own. Similarly, personalization systems can override individual choices, limiting the freedom for users to make their own shopping decisions—think of a “just for you” carousel that dominates an e-commerce homepage, with no way to customize preferences or remove unwanted items. 

These personalization attempts fall short because they focus on delivering highly personalized suggestions that make decisions for users rather than supporting individual autonomy. According to reactance theory, people naturally push back against these perceived impositions on their freedom.⁶ Telling the user what’s best for them or undermining their ability to make their own choices restricts their sense of control and threatens their autonomy. This often triggers emotional and cognitive processes that elicit behavior intended to restore autonomy, like dismissing recommendations or shopping elsewhere.

In the context of personalized advertising, autonomy isn’t just about making independent shopping choices. It can also refer to the user’s sense of control over their own data. When a recommendation system invades the user’s privacy—for example, it seems to know about a preference the user didn’t explicitly share—it can trigger the same defensive reaction. One study exploring why people install ad-blocking software found that psychological reactance to privacy invasion is a key driver behind this push-back behavior.⁶ This reaction was especially strong when users didn’t see much benefit from the personalized recommendation.

Overall, reactance is more likely when personalized advertising feels manipulative, out of touch, or one-sided, where it seems to benefit the company more than the consumer. This is why even well-intentioned personalization strategies can backfire when they overlook the user’s identity and goals.

Behavioral science tells us that product recommendations are less likely to trigger resistance when they feel more like extensions of the self than external impositions. Personalization should reflect real user values, goals, and identities, but also give users the option to make their own decisions. There are two main ingredients to include here: identity congruence and allusions to choice.

Identity Congruence Reduces Reactance to Persuasive Messaging

Identity congruence refers to the match between the identity of the consumer and the brand they’re engaging with.⁹ The more these identities overlap, the more brand messaging feels personally meaningful to consumers. The problem with a lot of personalization strategies is that they fail to match actual consumer identities because they rely on superficial browsing data to determine what customers want. As a result, consumers lose the chance to make purchase decisions that reflect their true goals and values.

Fortunately, research suggests that helping consumers identify with brands can reduce reactance to persuasive messaging and increase the chance that they’ll like the products being recommended. In one study, participants were allowed to customize an avatar to represent the source of persuasive messaging.¹⁰ Giving participants more options to customize the appearance of the avatar increased perceptions of identity alignment with the source, both reducing psychological reactance and increasing product liking. Other research shows that highly personalized messages are less likely to elicit reactance when the receiver perceives higher utility from the message.⁶ 

This research suggests that personalization is more likely to resonate when it feels relevant and useful to the customer. Sounds like a no-brainer, right? But how do you do this? Something as simple as allowing customers to shape their own user persona—for example, by selecting product preferences or describing their personal style—can increase identity alignment and improve goal matching. Asking consumers for their input ensures recommendations feel like they’re coming from a source that actually understands them, rather than an impersonal algorithm designed to analyze their behavior. Brands could also conduct diary studies to uncover the dynamic emotions, attitudes, and goals of digital product users to find out what factors make them more receptive to recommendations.

Allusions to Choice Extend a Sense of Control to Consumers

Giving users choices can also increase their perceived autonomy, further minimizing a resistant response to personalization.⁶ One way to do this is to do this by changing how the message itself is framed. For example, “Explore these product ideas” offers more autonomy than “These are the best options for you.” Another option is to change the physical environment in which the message is delivered. For instance, if a consumer receives a personalized product recommendation as a result of actively opting into a curated experience, they are less likely to feel that their freedom is being restricted or threatened. 
The key, according to behavioral research, is to avoid hyper-targeting messages to consumers based on deep layers of personal data or browsing behavior.⁶ This only makes consumers feel a loss of control over their information. To reduce privacy concerns and related reactance, promotional recommendations or personalized advertising should be born out of an established relationship with the brand, reflect a genuine understanding of the consumer, and extend some level of control to the user.

Another common issue with personalization systems is that many users simply don’t understand how they work. Recommendation algorithms are often seen as a kind of “black box.” It’s rarely made clear to users exactly what data is being used, for what purpose, whether or not it’s secure, and who is responsible for protecting it. Whether this lack of transparency is intentional or not, it degrades user trust in personalized recommendations.¹¹ 

Folk Theories Fill Gaps in Understanding

Why do users default to distrust when they don’t understand how algorithms work? Some researchers explain this through a concept in the field of human-computer interaction called folk theories.¹² Folk theories, specifically related to algorithms, are the everyday beliefs that people have about how these systems work, and they’re often implicit and imprecise. One common folk theory about algorithms is that they’re reductive, working based on stereotypes or simple assumptions. Another is that algorithms are exploitative, as in they’re primarily used to harvest personal data to manipulate consumers and rake in profits for companies. It’s not hard to see how these beliefs make it challenging for consumers to trust personalization algorithms. 

Users Rely on Heuristics to Determine Whether to Trust an Algorithm

When users lack detailed knowledge about algorithms, they also resort to heuristics to make judgments about whether these systems are trustworthy. Heuristics are mental shortcuts that allow people to make judgments and decisions quickly, often based on generalizations or “rules of thumb.” Research shows that users rely on four main heuristic cues when deciding whether to trust a recommender algorithm.¹¹ These cues include:

• Fairness: How well the algorithm avoids creating unfair bias or producing discriminatory results.
• Accountability: Whether the providers of the algorithm can be held accountable for unintended consequences, such as discrimination or possible harms that result from recommending certain products.
• Transparency: The extent to which recommendations made by algorithms are clear and obvious to users, such as what products are being recommended and why.
• Explainability: Whether the methods and techniques used by the personalization algorithm are understandable, offering insight into how the system arrives at certain outputs.

Fairness, accountability, transparency, and explainability (collectively known as FATE) are considered critical values in social systems, so it makes sense that people would use these same values to evaluate trust in algorithm-based services. In fact, users experience a higher level of trust in algorithms when they perceive higher levels of FATE characteristics.¹¹ When these signals are absent, however, users are more likely to feel that the system is watching them, unnecessarily prying into their behavior, or collecting their data for corporate gain.

The FATE framework isn’t just important for ensuring the ethical use of personalization technology, but also for giving users confidence to engage with these systems. Brands that prioritize FATE values can shift user perceptions and establish personalization as helpful rather than suspicious. Let’s explore how transparency cues and privacy nudges can make personalization algorithms feel less mysterious and malicious while improving users’ sense of control over their data.

Transparency Cues Can Reduce Uncertainty

Transparency cues are all about explaining the why behind product recommendations. When users feel that algorithms are making fair and rational choices, they’re more likely to see these systems as trustworthy and useful.¹¹ A transparency cue might be as simple as a message stating “Based on your recent searches” or “Since you added X product to your cart,” much like how Netflix prefaces its personalized collections with “Because you watched X.” Connecting the dots for consumers can help reduce some of their uncertainty about how personalization algorithms decide what to recommend.

Transparency cues can go a step further to increase perceived fairness and accountability of algorithms, combating the folk theory that personalization recommendations mainly make assumptions based on the user’s age, gender, or other traits. For example, a message stating, “These recommendations are based on your recently saved items, but let us know if we missed the mark” can reassure users that the goal is to be helpful rather than to persuade.

Privacy Nudges Improve User Understanding

When it comes to designing product recommendation systems, every design decision nudges users in one direction or another.¹³ Sometimes, these nudges push users toward risky behaviors, like giving apps unrestricted access to personal data. However, nudges can also direct users toward more protective behaviors, like only sharing necessary information.

While nudges take many forms, informational nudges can help users understand why personalization algorithms need access to user data so they feel confident making decisions about what to disclose and what to keep private. Examples include brief prompts that explain why certain data is being requested or progressive disclosure notices that outline privacy options as they come up, rather than burying important details in the fine print. Believe it or not, this won’t necessarily reduce the amount of data that can be collected and used for personalization. Research shows that users are more likely to approve data permission requests that include an explanation, suggesting that keeping the user in the loop can be beneficial for brands, too.¹³ 

One of the most important things about these informational nudges is that they’re easy to understand and act on. Giving users too much information or presenting them with too many settings all at once increases their cognitive load, reducing their ability to understand what’s going on and make confident decisions.¹³ The aim should be to support cognitive fluency and promote a sense of ease. To further reduce decision-making barriers, informational nudges can go hand in hand with default settings—for example, pre-selecting the least intrusive privacy settings instead of requiring that users actively select options that are in their best interest. When defaults are paired with clear explanations, they simplify the experience, helping users feel in control without being overwhelmed.

Clearly, personalization can go wrong when recommendations feel too controlling and algorithms lack transparency. But there’s another challenge that brands face when personalizing product recommendations—customers are hesitant to give up their data because the risks often outweigh the rewards. Consider that only 33% of consumers trust companies to use their data responsibly.² Consumers are increasingly aware of the use of third-party data by advertisers and the practice of brands selling customer data to third-party aggregators for profit.⁶ As a result, users often opt out of data collection entirely, even when brands have good intentions. The question now becomes: how can companies encourage customers to share their data willingly in the face of these perceived risks?

Consumers Are Motivated to Avoid Risk Over Gaining Rewards

Based on the theory of privacy calculus, deciding whether or not to share data largely comes down to an evaluation of risk vs. reward.¹⁴ Users weigh the perceived benefits of sharing their data against the potential costs, and when the risks outweigh the anticipated rewards, the choice becomes obvious—opt-out of data sharing and decline those permission requests. Even when rewards are about equivalent to the potential risks, users still tend to opt out of sharing to stay on the safe side. 

This aversion to risk is rooted in the concept of loss aversion, which explains how we rather avoid a potential loss than gain a reward of similar magnitude. The problem is that the benefits of personalization for the end-user are not always obvious. So, even when companies reassure users that the risks to their privacy are low, the lack of clear benefits tips the scale in the wrong direction.

Reciprocity Drives Consumer Willingness to Share Data

Researchers have also used social exchange theory to explore users’ willingness to share personal information and receive personalized recommendations.¹⁴ Social exchange theory asserts that social interactions, whether offline or in the digital world, are rooted in reciprocity, where individuals exchange things with the expectation of mutual benefit. In fact, users willingly disclose their data when they anticipate receiving benefits, whether these benefits include access to a platform, personalized experiences, or the use of free services. Almost half of consumers are willing to exchange their data for the simple reward of a better brand experience.¹⁵ 

However, the landscape of personalization systems is largely seen to benefit corporations more than the consumer, so users tend to underestimate the personal value they receive in exchange.¹⁴ This creates the perception of an unfair exchange, and why would you give up something valuable to receive something less valuable in return?

Beyond reducing perceived risk and improving user control, incentives can go a long way toward encouraging data sharing. When consumers give companies their data in anticipation of personal benefits, they do not experience the same reactive response to personalization that they might experience when the benefits are unclear.⁶ To incentivize data sharing, companies should focus on highlighting the concrete benefits users gain from personalized recommendations. 

Importantly, these benefits need to be immediate and tangible, not abstract promises like “Tell us about yourself so we can make your experience better.” For example, a pop-up on an online clothing store might say, “Discover the perfect fit for your personal style—just answer 3 quick questions.” This messaging reframes the data request as an offer of an exchange. The key here is to show the value of personalization before asking users for their data. Leading with the benefit and following up with the request leverages the anchoring effect, where people tend to rely on the first piece of information they encounter when making decisions. Encouraging users to focus on the reward rather than the cost can make them more receptive to data requests. 

Many brands already do this successfully, treating data collection like a social exchange rather than a one-sided request. For example, Sephora has a number of quizzes designed to help shoppers find products that align with their personal skincare goals, and many users appreciate these inquires because they make the shopping experience easier and less overwhelming. Likewise, mattress brand Casper offers a brief questionnaire to help shoppers find mattresses that support their sleep habits, goals, and budget. Whether it’s full quizzes or one-off questions, collecting high-quality data from consumers should be mutually beneficial.

Social Proof Reinforces Rewards

Social proof is a valuable tool for highlighting the rewards users can gain from sharing their data with brands. Social proof relies on our tendency to observe the actions of others to determine what to do, especially in uncertain situations.¹⁶ In the context of personalization technology, people’s willingness to engage with these tools is often shaped by the behavior of others. When users see other people interacting with personalized products, they are more likely to feel that the technology is beneficial. In one study on consumer resistance to AI-generated medical recommendations, researchers found that showcasing the number of satisfied customers helped enhance trust and willingness to engage with this technology.¹⁶ 

When it comes to product recommendations, this might mean sharing a simple stat reflecting the number of shoppers who found value in the recommendation. Of course, it’s crucial to follow through on the above promises by delivering real, immediate value to the user in exchange for their data, and only collecting the necessary data to personalize the experience.

There are a couple of important things to keep in mind when implementing behavioral design strategies intended to improve psychological privacy. The main issue here is the potential for these strategies to be misleading if they are not coupled with genuine measures to improve actual privacy. Otherwise, methods like transparency cues, social proof, identity alignment, incentives, and privacy nudges can be seen as performative and manipulative. Before working to build trust in recommendation algorithms, product teams should implement real safeguards to protect user privacy, such as only collecting data that is strictly necessary for delivering personalized recommendations, explicitly asking users for consent, investing in robust cybersecurity measures, and conducting audits to ensure compliance with regulations.

These measures go hand-in-hand with prioritizing zero-party data (data directly shared by customers) and first-party data (collected from customers’ behavior) rather than second-party data (acquired from brand partnerships) or third-party data (gathered and sold by external aggregators). Not only is zero- and first-party data higher quality and more reliable, but it’s also far less likely to elicit resistance from consumers.¹⁷ As regulators crack down on data sharing and selling—and consumers grow increasingly wary about these practices—a focus on high-quality data and robust privacy measures is key to creating personalized customer experiences that feel authentic and trustworthy.

Personalization and privacy don’t have to contradict each other. Paired with concrete methods to ensure data privacy, behavioral design strategies intended to elevate psychological privacy can ensure both brands and consumers benefit from personalized product recommendations. Giving users options can empower autonomy and reduce reactance to personalized suggestions, transparency cues coupled with privacy nudges can make consumers less skeptical about black box algorithms, and delivering real value to users can ensure the rewards outweigh the perceived risks. Each of these tools can help brands get closer to a personalization strategy that elicits engagement instead of backlash.

Moving from profit-focused product recommendations to value-based personalization is becoming increasingly important as global privacy regulations get stricter. The global movement to protect consumer data reflects a growing realization that users deserve control over how their data is used. As laws limit the data that businesses are allowed to collect, privacy-first personalization can give companies a significant strategic advantage.¹⁸ Not only that, but brands that embrace ethical and transparent personalization practices can lead by example, helping us move toward a global e-commerce environment where users feel confident sharing their preferences, shopping habits, and goals in exchange for better digital experiences.

Just like personalization, behavioral design is not a one-size-fits-all process. The best way to protect the psychological privacy of online users can vary widely across industries, cultures, and contexts. At The Decision Lab, we specialize in conducting holistic consumer research to answer deep questions about consumer decisions and designing dynamic personalization strategies that resonate on an individual level. Get in touch today so we can build a targeted personalization strategy that puts your user front and center.

The Personalization Paradox: Balancing Convenience and Privacy

Personalization has the potential to save time and reduce choice overload by streamlining our decisions. However, these benefits come with some problematic drawbacks. In this article, we take a deeper look at the personalization paradox and explain how this trendy tech can be a double-edged sword.

This is Personal: The Do's and Don'ts of Personalization in Tech

Most tech companies use some level of personalization to increase engagement, but this strategy can quickly backfire when personalized content feels more creepy than useful. To resonate rather than breed resistance, personalization algorithms must match consumer emotions, attitudes, goals, and personality traits. Give this piece a read to see what this might look like.

1. Stein, E., Robinson, K., Wolfer, A., Almeida, G., & Huang, W. (2025, January 30). Unlocking the next frontier of personalized marketing. McKinsey & Company. https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/unlocking-the-next-frontier-of-personalized-marketing 
2. Qualtrics. (2025). 2025 Global consumer trends report. https://www.qualtrics.com/ebooks-guides/customer-experience-trends/ 
3. Sweeney, E. (2018, February 21). 75% of consumers find many forms of marketing personalization creepy, new study says. Marketing Dive. https://www.marketingdive.com/news/75-of-consumers-find-many-forms-of-marketing-personalization-creepy-new-s/517488/ 
4. Twilio Segment. (2022). State of personalization report. https://segment.com/state-of-personalization-report-2022/ 
5. Fazlioglu, M. (2023, March 22). Most consumers want data privacy and will act to defend it. International Association of Privacy Professionals. https://iapp.org/news/a/most-consumers-want-data-privacy-and-will-act-to-defend-it 
6. Brinson, N. H., Eastin, M. S., & Cicchirillo, V. J. (2018). Reactance to Personalization: Understanding the Drivers Behind the Growth of Ad Blocking. Journal of Interactive Advertising, 18(2), 136–147. https://doi.org/10.1080/15252019.2018.1491350 
7. Chen, X., Sun, J., & Liu, H. (2022). Balancing web personalization and consumer privacy concerns: Mechanisms of consumer trust and reactance. Journal of Consumer Behaviour, 21(3), 572-582. https://doi.org/10.1002/cb.1947 
8. Karlovitch, S. (2024, September 27). Privacy laws raise questions for advertisers: Here’s what the numbers say. Marketing Dive. https://www.marketingdive.com/news/privacy-laws-raising-questions-advertisers-numbers-data-protection/728072/ 
9. Roggeveen, A. L., Grewal, D., Karsberg, J., Noble, S. M., Nordfält, J., Patrick, V. M., Schweiger, E., Soysal, G., Dillard, A., Cooper, N., & Olson, R. (2021). Forging meaningful consumer-brand relationships through creative merchandise offerings and innovative merchandising strategies. Journal of Retailing, 97(1), 81-98. https://doi.org/10.1016/j.jretai.2020.11.006 
10. Hanus, M. D., & Fox, J. (2017). Source Customization Reduces Psychological Reactance to a Persuasive Message via User Control and Identity Perceptions. Journal of Interactive Advertising, 17(1), 1–12. https://doi.org/10.1080/15252019.2017.1287023 
11. Shin, D. (2020). User Perceptions of Algorithmic Decisions in the Personalized AI System:Perceptual Evaluation of Fairness, Accountability, Transparency, and Explainability. Journal of Broadcasting & Electronic Media, 64(4), 541–565. https://doi.org/10.1080/08838151.2020.1843357 
12. Ytre-Arne, B., & Moe, H. (2020). Folk theories of algorithms: Understanding digital irritation. Media, Culture & Society, 43(5), 807-824. https://doi.org/10.1177/0163443720972314 
13. Acquisti, A., Adjerid, I., Balebako, R., Brandimarte, L., Cranor, L. F., Komanduri, S., ... & Wilson, S. (2017). Nudges for privacy and security: Understanding and assisting users’ choices online. ACM Computing Surveys (CSUR), 50(3), 1-41. https://doi.org/10.1145/3054926 
14. Cloarec, J., Meyer-Waarden, L., & Munzel, A. (2024). Transformative privacy calculus: Conceptualizing the personalization-privacy paradox on social media. Psychology & Marketing, 41(7), 1574-1596. https://doi.org/10.1002/mar.21998 
15. Jack Morton. (2024, September). Experience Better /Insights. https://jackmorton.com/wp-content/uploads/2024/09/EBI_PartOne_Final.pdf 
16. Alshakhsi, S., Almourad, M. B., Babkir, A., Al-Thani, D., Yankouskaya, A., Montag, C., & Ali, R. (2025). Designing AI to foster acceptance: do freedom to choose and social proof impact AI attitudes among British and Arab populations? Behaviour & Information Technology, 1–19. https://doi.org/10.1080/0144929X.2025.2477053 
17. Twilio Segment. (2021). State of personalization report. https://segment.com/state-of-personalization-report-2021/ 
18. Beohar, A. (2024, November 1). Inside the privacy-first approach to the personalized customer experience. CMSWire. https://www.cmswire.com/customer-experience/inside-the-privacy-first-approach-to-the-personalized-customer-experience/